accusative and ergative languages

Ralf-Stefan Georg Georg at home.ivm.de
Sat Jun 12 09:50:03 UTC 1999


>I do not think you are responding to what I am saying. Your examples suggest
>to me that you are under the impression that I denied that ergative
>structure could develop from accusative structure. I did not say that nor do
>I believe that. I do continue to believe that any language which is
>presently or has been accusative must have gone through an ergative stage
>sometime prior to that in its development.

>For most transitive verbs, I believe the closest connection is between it
>and its object so that, at some stage of development, an unmarked verb form
>should represent a passive. To try to understand ergative constructions from
>passive inflections developed in languages in accusative stages seems to me
>to be potentially misleading.

You said twice that this is something which you *believe*, and this is OK
with me, assuming that you are living in a community where you are free to
believe whatever you choose to.
But: this is of course without consequences for what we *know* about
ergativity.
First: as I think has been made clear several times here during this
thread: there is no sense in the notion "ergative language", since
ergativity is a phenomenon which may be present in some subsystems of a
given language and absent in others, thus, while definitely showing this
phenomenon to certain degrees, Georgian, Basque, Dyirbal and Thakali may be
called "ergative languages", butt hey are in sometimes very different ways.
What, in this respect, "going through an ergative stage" means is rather
equivocal and requires special definitions, which seem to be lacking in
your theory. "Ergative language" is a squishy notion.
Secondly: if we take every possible manifestation of ergativity into
account, we know hosts of cases where (in one, some, most subsystems of a
language) a) ACC > ERG and b) ERG > ACC, as well as both construction types
remaining stable through the observable history of a language. Logically,
from this follows nothing more that both types can precede the other, and
nothing more. The alleged necessity for ACC to be preceded by ERG at any
rate is just not contradicted by *this* fact, but that is all one can
derive from this. That, however, your scenario is the less likely one is
confirmed by some simple observations, namely that "ergative languages"
(used with the qualifications mentioned above, and loosely meaning a
language with a marked dominance of ergative constructions in various
subsystems, among them the alignment of the basic constituents in an
unmarked transitive sentence [note that even the superficially simple term
"transitivity" needs some definition before we can use it with sense;
however, instead of doing this here, I may refer you to Hopper/Thompson in
Lg 1980]) are globally in the minority, that they further tend to cluster
in specifiable regions (I know Basque is an exception, no need to remind
me), iow. that it is a phenomenon tending to areal spread aso. Furthermore,
and what may be more significant, while we do know a great deal of
languages without a single discernable trait of it, i.e. fully ACC
languages, fully ERG languages don't seem to exist, i.e. all known
languages with some ergativity display at least one subsystem which is
organized on an ACC basis, the reverse not being observable (iow: there are
only split-ergative languages, admittedly sometimes with less salient
splits, but never without one). All this makes ergativity a  more *marked*
construction type than accusativity. This does not rule out the possibility
of languages, for which only full accusativity is observable throughout
their attested or confidently reconstructable history may not have had a
(more) ERG past nevertheless, but in the absence of conclusive and specific
evidence for this (the precise nature of which can, of course, be
discussed) there can be no automatic rule which would force us to assume
this.
Finally, as Larry Trask has pointed out before, the idea of stadiality  in
language change has been safely laid to rest long ago, being nothing less
than aprioristic ideology (if you like to stick to it nevertheless, you
should be aware that you are in the fine company of, among others, N.Ja.
Marr). It is no better confirmed than the notion that, e.g., feudalism
precedes capitalism, which will yield to socialism eventually, or, back to
linguistics, that all languages which display mostly pulmonic consonants,
must have been preceded of necessity by stages which displayed a dominance
of clicks in their systems.
The idea of stadialism in language - which is of course a social
institution - is philosophically on the same level as any other theory
which tries to subject social institutions to inalterable laws of
teleological development, like that of historical materialism.

St.G.

Stefan Georg
Heerstrasse 7
D-53111 Bonn
FRG
+49-228-69-13-32



More information about the Indo-european mailing list